pr-checks（公关检查）

PR Checks Skill

This skill runs comprehensive PR checks to ensure code quality and address review feedback on an existing pull request.

When to Use

Run this skill when:

• A PR has been created and CodeRabbit has posted review comments
• You want to address existing review feedback systematically
• Before requesting final review/merge on a PR

Usage

/pr-checks

Workflow

1. Analyze Current State

First, understand what's being reviewed:

# Check current branch and status
git branch --show-current
git status --short

# Show diff summary vs main
git diff --stat main...HEAD

# Get the PR number for this branch
gh pr view --json number,title,state

2. Review and Fix PR Description

Ensure the PR has a proper description before addressing code comments:

# View current PR description
gh pr view --json body,title

A good PR description should include:

• Summary: 1-3 bullet points explaining what the PR does
• Test plan: How to verify the changes work
• Related issues: Links to Linear/GitHub issues (e.g., Closes MAS-XXX)

If the description is missing or inadequate:

# Update the PR description
gh pr edit --body "$(cat <<'EOF'
## Summary
<1-2 sentence overview of what this PR accomplishes>

### Changes
- <change 1: what was added/modified/removed>
- <change 2>
- <change 3>
- ...

### Technical details (if applicable)
<Brief explanation of implementation approach, architectural decisions, or non-obvious changes>

## Test plan
- [ ] <verification step 1>
- [ ] <verification step 2>
- [ ] <edge case or error scenario tested>

## Related issues
Closes MAS-XXX

## Screenshots/recordings (if applicable)
<Add screenshots for UI changes, terminal output for CLI changes>

🤖 Generated with [Claude Code](https://claude.com/claude-code)
EOF
)"

Fix the title if needed (should follow conventional commits format):

# Update PR title
gh pr edit --title "feat: descriptive title here"

3. Review Existing CodeRabbit Comments

This is the primary workflow. CodeRabbit automatically reviews PRs and posts comments. Use /pr-comments to fetch and process them:

/pr-comments

This fetches all review comments from the PR. For each CodeRabbit comment:

1. Read the comment - Understand what CodeRabbit is suggesting
2. Evaluate relevance - Is this a valid concern for this codebase?
3. Decide action - One of:

- ✅ Implement - The suggestion is valid and worth fixing - ❌ Skip - The suggestion doesn't apply or is too minor - ❓ Clarify - Need more context from user before deciding

Walk through comments one by one with the user:

## CodeRabbit Comment #1 of 5

**File:** `massgen/backend/foo.py:45`

**Original code:**

response = client.api_call(params) return response.data

**CodeRabbit suggestion:**

> Consider adding error handling for the API call. The request could fail due to network issues or API errors, which would cause an unhandled exception. This is especially important since this is called from the main orchestration loop where failures could crash the entire run. [truncated - 8 more lines]

**Suggested change:**

try: response = client.api_call(params) return response.data except APIError as e: logger.error(f"API call failed: {e}") raise

**My assessment:** Valid concern - the API call could fail and there's no error handling.

**Recommendation:** ✅ Implement

Do you want me to:

1. Implement this fix
2. Skip this comment
3. Need more information

After user decides, resolve the comment on GitHub:

# If implemented or intentionally skipped, resolve the comment thread
gh api graphql -f query='
  mutation {
    resolveReviewThread(input: {threadId: "THREAD_ID"}) {
      thread { isResolved }
    }
  }
'

Alternatively, reply to the comment explaining the action taken:

# Reply to the comment
gh pr comment <PR_NUMBER> --body "Addressed in <commit-sha>: <brief description of fix>"

When showing comments:

• Show the original code being discussed
• Show the full suggestion text (truncate if >15 lines with "[truncated - N more lines]")
• Show the suggested change if CodeRabbit provided one
• Include line numbers for context

Wait for user approval before implementing each fix. This ensures:

• User maintains control over what changes are made
• No unnecessary changes are introduced
• Context-specific decisions can be made

4. Run Pre-commit Hooks

After making fixes, run pre-commit to ensure code style:

uv run pre-commit run --all-files

If issues are found, fix them and commit.

5. Run Tests

# Run tests (skip expensive API tests)
uv run pytest massgen/tests/ -v -m "not expensive and not docker" -x --tb=short

6. Validate Configs (if modified)

uv run python scripts/validate_all_configs.py

7. Commit and Push Fixes

# Stage fixes
git add -u .

# Commit with descriptive message
git commit -m "fix: address CodeRabbit review comments

- Fix error handling in foo.py
- Add missing type hints in bar.py

🤖 Generated with [Claude Code](https://claude.com/claude-code)

Co-Authored-By: Claude <noreply@anthropic.com>"

# Push to trigger CodeRabbit re-review
git push

8. Run PR Review Toolkit (Optional)

For additional analysis beyond CodeRabbit:

/pr-review-toolkit:review-pr

This runs specialized agents for:

• Code review against project guidelines
• Silent failure detection
• Type design analysis
• Test coverage analysis

9. (Optional) Run Local CodeRabbit Review

If you want to run a fresh local review (separate from GitHub PR comments):

coderabbit --prompt-only --type committed

Note: This runs CodeRabbit's own analysis locally, which may differ from the automated PR review. The GitHub PR comments from step 3 are typically more thorough since they have full PR context.

10. Generate Summary

After all checks, output a summary:

## PR Checks Summary

### Branch: feature/my-feature
- PR #123: "feat: add new feature"
- Commits ahead of main: 3
- Files changed: 5

### PR Description
✅ Has summary, test plan, and issue links

### CodeRabbit Comments Addressed

| Comment | File | Action |
|---------|------|--------|
| Add error handling | foo.py:45 | ✅ Implemented |
| Consider caching | bar.py:12 | ❌ Skipped (not applicable) |
| Missing type hint | baz.py:78 | ✅ Implemented |

### Check Results

| Check | Status |
|-------|--------|
| Pre-commit | ✅ Passed |
| Tests | ✅ 47 passed |
| Config Validation | ✅ Valid |

### Ready for Merge?
✅ All critical issues addressed, ready for final review

Reference

PR Description Template

## Summary
<1-2 sentence overview of what this PR accomplishes>

### Changes
- <change 1: what was added/modified/removed>
- <change 2>
- <change 3>

### Technical details (if applicable)
<Implementation approach, architectural decisions, or non-obvious changes>

## Test plan
- [ ] Step to verify functionality
- [ ] Edge cases tested
- [ ] Error scenarios handled

## Related issues
Closes MAS-XXX

## Screenshots/recordings (if applicable)
<For UI/CLI changes>

## Breaking changes (if applicable)
<What breaks and migration steps>

PR Title Format

Use conventional commits format:

• feat: - New feature
• fix: - Bug fix
• docs: - Documentation only
• refactor: - Code change that neither fixes a bug nor adds a feature
• perf: - Performance improvement
• test: - Adding or updating tests
• chore: - Maintenance tasks

Pre-commit Hooks

• black - Python formatter (200 char line)
• isort - Import sorter
• flake8 - Style checker
• autoflake - Remove unused imports

Test Markers

• @pytest.mark.expensive - Skip in quick checks
• @pytest.mark.docker - Docker-dependent

CodeRabbit Config

See .coderabbit.yaml for path-specific review instructions and exclusions.